The purpose of this study is to examine the effects on patients, families and nurses of a new way of caring for chronically critically ill patients comparing those cared for in a nurse-managed Special Care Unit (SCU) to those cared for in traditional intensive care units (ICU's). Specific aims are: 1) to compare the effect of the Special Care Unit practice model with the traditional ICU environment on the satisfaction and retention of critical care nurses, and 2) to compare the effects of the SCU with the traditional ICU environment on patient outcomes and family satisfaction with care. The independent variable is type of ICU care (traditional ICU's versus nurse-managed SCU). Dependent variables are nurse outcomes of satisfaction, turnover and absenteeism, and patient outcomes of length of stay, mortality, respiratory and nosocomial infection complications, readmission rate, cost of care and patient and family satisfaction with care. This is a quasi-experimental design with random assignment of patients to two types of ICU care. Nurse assignment by necessity will be predetermined by unit of employment. A sample of 300 patients and 68 nurses is projected. Data on patient subjects will be collected daily while in the ICU and at the time of discharge for patient and family satisfaction. Data on nurse outcomes will be collected yearly with turnover and absenteeism monitored weekly throughout the project. Instruments include APACHE II, TISS, Nursing Complexity Index, Lamonica/Oberst Satisfaction Scale, Job Descriptive Index, Respiratory Complication Index and the Nosocomial Infection Index. Data analysis will consist of comparisons of the two groups of patient and nurse subjects on each measured outcome. The patient outcome variables of length of stay, number of complications, patient and family satisfaction and cost will be analyzed by use of a t-test for independent means. The variables of mortality and readmission rate will be analyzed by use of a chi-square statistic. The nurse outcome variables of job satisfaction and absenteeism will be analyzed by use of an ANOVA for unequal groups. The variable of turnover will be analyzed by use of a chi-square. Further analysis of patient outcomes will be done controlling for patient's age, sex, nursing complexity (NCI), acuity (APACHE II) number of interventions required (TISS), and for nurse outcomes controlling for age education, and length of experience. The development of this SCU provides a unique opportunity to test an innovative model for delivery of care from the perspective of the care givers (nurses) and care recipients (patients and families).